7 |
=cut |
=cut |
8 |
|
|
9 |
use strict; |
use strict; |
10 |
use Data::Dumper; |
use locale; |
11 |
|
use YAML; |
12 |
use Carp; |
use Carp; |
13 |
use Getopt::Long; |
use Getopt::Long; |
14 |
|
use Text::Unaccent 1.02; |
15 |
|
|
16 |
use lib './lib'; |
use lib './lib'; |
17 |
use WebPAC; |
use WebPAC; |
33 |
"low_mem!" => \$low_mem, |
"low_mem!" => \$low_mem, |
34 |
); |
); |
35 |
|
|
36 |
|
my $filter = { |
37 |
|
'CROVOC' => sub { |
38 |
|
my $tmp = shift || return; |
39 |
|
$tmp =~ s/\s*HIDRA.*//; |
40 |
|
return undef unless ($tmp =~ s/\s*CROVOC.*$/ #C#/); |
41 |
|
# remove repeating stars |
42 |
|
# FIXME this should be fixed at right place, not here! |
43 |
|
$tmp =~ s/(\s*#C#)+/ #C#/g; |
44 |
|
return $tmp; |
45 |
|
}, |
46 |
|
'CROVOC_tree' => sub { |
47 |
|
my $tmp = shift || return; |
48 |
|
$tmp =~ s/\s*CROVOC.*$/ <img src="img\/crovoc.png" border="0">/; |
49 |
|
$tmp =~ s/\s*EUROVOC.*//; |
50 |
|
$tmp =~ s/\s*HIDRA.*//; |
51 |
|
return $tmp; |
52 |
|
}, |
53 |
|
# TT filter |
54 |
|
'CROVOC_img' => sub { |
55 |
|
my $tmp = shift; |
56 |
|
$tmp =~ s/\s*#C#\s*/ <img src="..\/img\/crovoc.png" border="0">/gis; |
57 |
|
$tmp =~ s/"img\/crovoc.png"/"..\/img\/crovoc.png"/gis; |
58 |
|
return $tmp; |
59 |
|
} |
60 |
|
}; |
61 |
|
|
62 |
|
## remove accented characters |
63 |
|
# |
64 |
|
sub unac { |
65 |
|
my $string = shift || return; |
66 |
|
$string = unac_string($code_page,$string); |
67 |
|
$string =~ tr/đĐ/dD/; |
68 |
|
$string = unac_entities($string); |
69 |
|
return $string; |
70 |
|
} |
71 |
|
sub unac_2 { |
72 |
|
my $string = shift || return; |
73 |
|
if (length($string) > 2) { |
74 |
|
my $pr = substr($string,0,2); |
75 |
|
$string = unac_string($code_page,substr($string,2)); |
76 |
|
$string =~ tr/đĐ/dD/; |
77 |
|
$string = $pr . $string; |
78 |
|
} |
79 |
|
$string = unac_entities($string); |
80 |
|
return lc($string); |
81 |
|
} |
82 |
|
|
83 |
|
sub unac_entities { |
84 |
|
my $ent = shift || return; |
85 |
|
|
86 |
|
$ent =~ s/&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);/$1/gi; |
87 |
|
$ent =~ s/ð/d/g; |
88 |
|
$ent =~ s/&E[tT][hH];/D/g; |
89 |
|
$ent =~ s/&(\w\w)lig;/$1/gi; |
90 |
|
|
91 |
|
return $ent; |
92 |
|
} |
93 |
|
|
94 |
# create WebPAC object |
# create WebPAC object |
95 |
# |
# |
96 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
99 |
start_mfn => $start_mfn, |
start_mfn => $start_mfn, |
100 |
debug => $debug, |
debug => $debug, |
101 |
low_mem => $low_mem, |
low_mem => $low_mem, |
102 |
|
filter => $filter, |
103 |
) || die; |
) || die; |
104 |
|
|
105 |
my $log = $webpac->_get_logger() || die "can't get logger"; |
my $log = $webpac->_get_logger() || die "can't get logger"; |
108 |
|
|
109 |
my $index = new WebPAC::jsFind( |
my $index = new WebPAC::jsFind( |
110 |
index_path => $index_path, |
index_path => $index_path, |
111 |
keys => 10, |
keys => 62, |
112 |
) || die; |
) || die; |
113 |
|
|
114 |
my $thes; |
my $thes; |
118 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
119 |
filename => shift @ARGV || '/data/hidra/THS/THS', |
filename => shift @ARGV || '/data/hidra/THS/THS', |
120 |
lookup => [ |
lookup => [ |
121 |
{ 'key' => 'd:v900', 'val' => 'v250^a' }, |
{ 'key' => 'd:v900', 'val' => 'filter{CROVOC_tree}v250^a v800' }, |
122 |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Područje"', 'key' => 'pa:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
123 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
124 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
125 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
126 |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
127 |
# tree structure |
# tree structure |
128 |
{ 'eval' => 'length("v251") == 2', 'key' => 'root:v251', 'val' => 'v900' }, |
{ 'eval' => 'length("v251") == 2 && "v800" =~ m/EUROVOC/ || "v800" =~ m/CROVOC/ && "v251" =~ m/^(H|HD|L|Z|P)$/', 'key' => 'root:v251', 'val' => 'v900' }, |
129 |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
{ 'eval' => '"v251"', 'key' => 'code:v900', 'val' => 'v561^4:v251' }, |
130 |
|
{ 'eval' => '"v561^4" && "v562^4"', 'key' => 'code:v900', 'val' => 'v561^4:v562^4' }, |
131 |
], |
], |
132 |
); |
); |
133 |
|
|
144 |
my @ds = $webpac->data_structure($rec); |
my @ds = $webpac->data_structure($rec); |
145 |
|
|
146 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
147 |
$log->debug("rec = ",Dumper($rec)); |
$log->debug("rec = ",Dump($rec)); |
148 |
$log->debug("ds = ",Dumper(\@ds)); |
$log->debug("ds = ",Dump(\@ds)); |
149 |
} |
} |
150 |
|
|
151 |
next if (! @ds); |
next if (! @ds); |
176 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
177 |
next if (! $ds->{'swish'}); |
next if (! $ds->{'swish'}); |
178 |
|
|
179 |
|
# strip all non word characters from beginning or end |
180 |
|
# of word |
181 |
|
my $words = unac(join(" ",@{$ds->{'swish'}})); |
182 |
|
$words =~ s/^\W+//; |
183 |
|
$words =~ s/\W*\s+\W*/ /g; |
184 |
|
$words =~ s/\W+$//; |
185 |
|
|
186 |
|
# first try to generate headline for this entry from index |
187 |
|
my $h = $ds->{'index'}->[0]; |
188 |
|
# then, from display |
189 |
|
$h ||= $ds->{'display'}->[0]; |
190 |
|
# and as last resport, fallback to headline |
191 |
|
$h ||= $headline; |
192 |
|
|
193 |
$index->insert( |
$index->insert( |
194 |
index_name => $ds->{'tag'}, |
index_name => $ds->{'tag'}, |
195 |
path => $f, |
#path => $f, |
196 |
headline => $headline, |
path => $webpac->mfn, |
197 |
words => join(" ",@{$ds->{'swish'}}) |
headline => $h, |
198 |
|
words => $words, |
199 |
); |
); |
200 |
} |
} |
201 |
|
|
203 |
foreach my $ds (@ds) { |
foreach my $ds (@ds) { |
204 |
next if (! $ds->{'index'}); |
next if (! $ds->{'index'}); |
205 |
|
|
206 |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
$thes->{$ds->{'tag'}} ||= new WebPAC::Index( name => $ds->{'tag'} ); |
207 |
|
|
208 |
foreach my $h (@{$ds->{'index'}}) { |
foreach my $h (@{$ds->{'index'}}) { |
209 |
$thes->{$ds->{'tag'}}->insert( |
$thes->{$ds->{'tag'}}->insert( |
210 |
path => $f, |
sort_by => unac_2($h), |
211 |
|
mfn => $webpac->mfn, |
212 |
headline => $h, |
headline => $h, |
213 |
); |
); |
214 |
} |
} |
215 |
} |
} |
216 |
|
|
217 |
# print Dumper(\@ds); |
# print Dump(\@ds); |
218 |
|
|
219 |
} |
} |
220 |
|
|
238 |
} |
} |
239 |
|
|
240 |
if (0 && $log->is_debug) { |
if (0 && $log->is_debug) { |
241 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
242 |
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
$log->debug("data hash: ",Dump($webpac->{'data'})); |
243 |
foreach my $t (keys %{$thes}) { |
foreach my $t (keys %{$thes}) { |
244 |
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
$log->debug("thesaurus $t hash: ",Dump($thes->{$t})); |
245 |
} |
} |
246 |
} |
} |
247 |
|
|
248 |
} # XXX if(0) |
} # XXX if(0) |
249 |
|
|
250 |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
#$log->debug("lookup hash: ",Dump($webpac->{'lookup'})); |
251 |
|
|
252 |
$log->info("creating tree"); |
$log->info("creating tree"); |
253 |
|
|
267 |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"root:".$_[0]}} }, |
268 |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[1]}} }, |
269 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[1]}} }, |
270 |
have_children => sub { defined($l->{"a:".$_[0]."::"}) }, |
have_children => sub { return $l->{"a:".$_[0]."::"} }, |
271 |
child_code => sub { return $_[0] }, |
have_children_at_level => sub { |
272 |
|
return unless (defined($l->{"code:".$_[1]})); |
273 |
|
my $code = shift @{$l->{"code:".$_[1]}}; |
274 |
|
print STDERR "## $_[1] -> $code\n"; |
275 |
|
return undef unless($code); |
276 |
|
return(9, $l->{"a:$code:"} ) if (defined($l->{"a:$code:"})); |
277 |
|
}, |
278 |
style => 'display: none', |
style => 'display: none', |
279 |
|
iframe => 1, |
280 |
},{ |
},{ |
281 |
# 1 |
# 1 |
282 |
code_arr => sub { @{$l->{"a:".$_[0]."::"}} }, |
code_arr => sub { @{$_[0]} }, |
283 |
filter_code => sub { shift }, # nop |
filter_code => sub { shift }, # nop |
284 |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
285 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
286 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
287 |
have_children => sub { defined($l->{"a:".$_[1].":"}) }, |
have_children => sub { return $l->{"a:".$_[1].":"} }, |
|
child_code => sub { return $_[1] }, |
|
288 |
style => 'display: none', |
style => 'display: none', |
289 |
},{ |
},{ |
290 |
# 2 |
# 2 |
291 |
code_arr => sub { @{$l->{"a:".$_[0].":"}} }, |
code_arr => sub { @{$_[0]} }, |
292 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
293 |
lookup_v900 => sub { shift }, |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
294 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
295 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
296 |
have_children => sub { defined($l->{"a:".$_[2].":".$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return "a:".$_[2].":".$_[1] }, |
|
297 |
style => 'display: none', |
style => 'display: none', |
298 |
},{ |
},{ |
299 |
# 3 uži pojam |
# 3 uži pojam |
300 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
301 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
302 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
303 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
304 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
305 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
306 |
},{ |
},{ |
307 |
# 4 |
# 4 |
308 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
309 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
310 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
311 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
312 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
313 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
314 |
},{ |
},{ |
315 |
# 5 |
# 5 |
316 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
317 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
318 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
319 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
320 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
321 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
322 |
},{ |
},{ |
323 |
# 6 |
# 6 |
324 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
325 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
326 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
327 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
328 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
329 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
330 |
},{ |
},{ |
331 |
# 7 |
# 7 |
332 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
333 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
334 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
|
my ($c,$p) = @_; |
|
|
$p =~ s/^a:(..:....):.*$/$1/; |
|
|
return "a:".$p.":".$c; |
|
|
}, |
|
335 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
336 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
337 |
have_children => sub { defined($l->{$_[1]}) }, |
have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
|
child_code => sub { return $_[1] }, |
|
338 |
},{ |
},{ |
339 |
# 8 |
# 8 |
340 |
code_arr => sub { @{$l->{$_[0]}} }, |
code_arr => sub { @{$_[0]} }, |
341 |
filter_code => sub { shift }, |
filter_code => sub { shift }, |
342 |
lookup_v900 => sub { |
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
343 |
my ($c,$p) = @_; |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
344 |
$p =~ s/^a:(..:....):.*$/$1/; |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
345 |
return "a:".$p.":".$c; |
# have_children => sub { return $l->{"a:".$_[1].":".$_[0]} }, |
346 |
}, |
have_children => sub { 0 }, |
347 |
|
},{ |
348 |
|
# 9 - level which is never reached except explicitly |
349 |
|
code_arr => sub { @{$_[0]} }, |
350 |
|
filter_code => sub { shift }, |
351 |
|
lookup_v900 => sub { shift @{$l->{"code:".$_[0]}} }, |
352 |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
lookup_term => sub { shift @{$l->{"d:".$_[0]}} }, |
353 |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
lookup_mfn => sub { shift @{$l->{"900_mfn:".$_[0]}} }, |
|
# have_children => sub { defined($l->{$_[1]}) }, |
|
|
# child_code => sub { return $_[1] }, |
|
354 |
have_children => sub { 0 }, |
have_children => sub { 0 }, |
355 |
child_code => sub { 0 }, |
have_children_at_level => sub { defined($l->{"a:".$_[1].":".$_[0]}) && return (9,$l->{"a:".$_[1].":".$_[0]}) }, |
356 |
|
},{ |
357 |
}); |
}); |
358 |
|
|
359 |
my $tree = new WebPAC::Tree( |
my $tree = new WebPAC::Tree( |
|
dir => './out', |
|
|
html => 'browse.html', |
|
|
template => './output_template/tree.tt', |
|
|
js => 'tree-ids.js', |
|
360 |
tree => \@tree, |
tree => \@tree, |
361 |
|
detail_url => sub { |
362 |
|
my $mfn = shift; |
363 |
|
my $url = "thes/${mfn}.html"; |
364 |
|
return $url if (-e "./out/$url"); |
365 |
|
}, |
366 |
|
nodes_dir => 'nodes', |
367 |
|
); |
368 |
|
|
369 |
|
$tree->output( |
370 |
|
dir => './out', |
371 |
|
html => 'browse.html', |
372 |
|
template_dir => './output_template/', |
373 |
|
template_tree => 'tree.tt', |
374 |
|
template_node => 'node.tt', |
375 |
|
js => 'tree-ids.js', |
376 |
); |
); |
377 |
|
|
378 |
|
$tree->output( |
379 |
|
dir => './eurovoc', |
380 |
|
html => 'hijerarhija.html', |
381 |
|
template_dir => './output_template/', |
382 |
|
template_tree => 'hijerarhija.tt', |
383 |
|
template_node => 'hijerarhija_node.tt', |
384 |
|
js => 'tree-ids.js', |
385 |
|
); |
386 |
|
|
387 |
|
|
388 |
$log->info("closing index"); |
$log->info("closing index"); |
389 |
$index->close; |
$index->close; |
390 |
|
|